WIT Press


Optimal Design Of Power Supply Systems Using Genetic Algorithms

Price

Free (open access)

Paper DOI

10.2495/CR080391

Volume

103

Pages

10

Page Range

391 - 400

Published

2008

Size

406 kb

Author(s)

J. R. Jimenez-Octavio & E. Pilo

Abstract

This paper presents an optimization model based on genetic algorithms for designing traction power supply systems. The proposed model is suitable both for planning new lines and for expanding the old ones, resulting in a more efficient operation as well as in lower investment costs. Minimization of fixed installation represents the optimization criterion for searching innovative designs that fulfil certain technical constraints: maximum voltage drops and maximum power consumption in the substations. The variables involved in the optimization problem are: number, type and location of railway overhead lines; and number, size and location of traction substations. Finally, the evaluation of possible designs involves simplified electrical modelling of the studied railway stretch. Thus, electrical simulations and calculations have been also adapted for their implementation in the genetic algorithm. A section of the Madrid-Barcelona high-speed line has been considered as a study case in order to analyze the performance of the proposed model. Results reveal the suitability of the new designs obtained with the presented model and its goodness and robustness. Keywords: railway, power supply system, optimization, genetic algorithms. 1 Introduction The design of the electrification is a complex task that involves different interdisciplinary analysis. Normally, the design process is done iteratively by refining candidate designs based on its estimated performance. In this process simulation tools are crucial to evaluate the performance in many different situations. However, they do not usually include criteria to modify candidate designs in order to obtain the final solution. Thus, the designer has to decide the

Keywords

railway, power supply system, optimization, genetic algorithms.